HUMAN ACTION RECOGNITION SYSTEM

Abstract

Human action identification is the major part of research area in computer science and its applications.  The main purpose of the human action identification is to find out the ongoing occurrences and their related data from captured video. This technique is mainly used for monitoring sick persons and many numbers of systems that is directly connected with the computer interface system.  Most of the real times application needs to recognize high level actions.  The human actions are divided into three types based on their complexity.  The first type is called as single performer actions.  Walking and bending are the examples of single performer actions.  The second type of human action is interaction. The interactions may be happened between human to human or human to object.  Punching and lift the bags are the examples of interaction type.   The third type of human action is group events.  Dance with group members is one of the examples of group action.   The camera was already installed in common places.  The human action recognition is used to find the actions of the peoples to identify the important things.  In this proposed system is used to identify the human action by using deep motion mapping concept and various computing algorithms.  Finally the accuracy levels of the machine learning algorithms are compared.